期刊名称：Archiwum Fotogrametrii, Kartografii i Teledetekcji
印刷版ISSN：2083-2214
出版年度：2007
卷号：17
语种：English
出版社：Main Board of Association of Polish Surveyors
摘要：Very high resolution satellite images with their valuable photo-interpretation properties and potential high level of geometric accuracy of end products are considered in many applications as a crucial source of data and information about the Earth surface. It is of special importance for such applications in which the area of interest is characterised by complex spatial structure. Growing needs for obtaining diverse categories of information about Earth surface,including land cover,require effective and efficient methods of digital analysis to be worked out. In contrast to traditional methods of digital classification,which use pixel as a reference unit of processing and the frame of discriminating features is basically defined in spectral space,the object-based approach allows to increase the set of discriminating features,including new elements related to texture,size,shape, widely understood neighbourhood,and context. Object-based approach often allows,to a large extent, to avoid some problems of traditional pixel-based classifiers,which result from high level of heterogeneity of identified areas and the frequent presence of so-called statistical noise,which is considered as a consequence of high spatial resolution. The finally created and identified objects in object-based analysis,in their spatial distribution form more natural image of reality. In the present paper,certain results are presented from a range of digital analysis of VHR satellite image,where the main goal was to achieve land cover data applying pixel and object-based approach. In first case,the supervised approach was used with known traditional algorithms. The object-based approach was adopted based on Definiens Professional set of tools. In the latter approach,the essential was also to include certain elements of knowledge and ancillary information with the aim to improve efficiency of the method and accuracy of final results. Given tests were performed for the terrain of diverse levels of spatial complexity. For the rural areas, an important issue was also to recognize the crops. The results showed (with the certain input assumptions) the positive aspects and limitations of applied approaches and methods,pointing at some visible advantages of the object-based approach.
其他摘要：Zdjęcia satelitarne o bardzo wysokiej rozdzielczości przestrzennej (VHR) stają się niezastąpionym w wielu zastosowaniach źródłem danych i informacji o powierzchni Ziemi ze względu na wysokie walory interpretacyjne i możliwe do uzyskania dokładności kartom
